1
Real-Time Adverse Event Predictive
Model Reduces Readmissions
Session #185, February 13, 2019
Dr. Manjula Julka, Vice President Clinical Innovation, PCCI
Kristin Alvarez, Associate Pharmacy Director, Parkland Hospital
2
Manjula Julka, MD, MBA
Has no real or apparent conflicts of interest to report.
Kristin Alvarez, PharmD, BCPS
Has no real or apparent conflicts of interest to report.
Conflict of Interest
3
• Define adverse drug event and its impact
• Summarize the concept of pharmacy-led interventions
• Illustrate design thinking framework and end user collaboration
Learning Objectives
4
National Burden - Readmissions
Private Pay:
$785M
Medicaid:
$839M
Medicare:
$4.3B
Readmissions
$41.3B
5
What is an ADE?
3rd
leading
cause of
death
Increases
LOS by
2-3 Days
Increases cost
by
$16K-$24K
Increases risk
for
readmission
Any harm or injury that
results from medication use.
ADE
INCIDENCE
Adverse Drug Events cost
$3.5B annually
6
Pharmacy-led Interventions are more likely
to...
58-66%
Reduction Med
Discrepancies
Lower preventable
ADEs
(1.76 times)
Cost less
(Pharm Tech/Pharm
Student)
Reduce ED
Visits/Readmissions
Identify Errors
(4.5 times)
7
Parkland Hospital
• 870-bed acute care Level hospital 1 Trauma Center
• Ambulatory Care
o 12 community-based clinics
o 12 school-based clinics
o 120 specialty clinics
o Mobile units
• Dallas County Jail health care
• Over 10.5 million prescriptions (FY2017)
• 250,000+ ED visits
8
Parkland Center for Clinical Innovation
AGILE ⃘ DESIGN THINKING ⃘ INNOVATION ⃘ COLLABORATION
ADVANCED DATA ARCHITECTURE & AI PLATFORM
MISSION: Reimagine and expand the knowledge base of
healthcare through prescriptive analytics and artificial intelligence to
deliver precision medicine.
9
Human-Centered Design Innovation
DISCOVER
DESIGN DELIVER
10
EMPATHIZE DEFINE IDEATE
PROTOTYPE
TEST
Design Thinking Framework
BUILD
MEASURE SUSTAIN
DESIGN DELIVERDISCOVER
11
Scope of Problem
274,198 ED visits - 74,977 Discharges
FY2017
EMPATHIZE
3 FTEs
Time and Resources
12
Discover
Inpatient
Workflow
Pharmacy-led
timely interventions
Improved
Outcomes
Patients admitted
to hospital
Who is at risk?
Who is at risk?
13
Design
IDEATE
DEFINE
14
Demographics
Medications
Clinical Data
Social
Determinants of
Health
Real-time
Predictive
Score
Model inputs
Dashboard
(Director view)
Flowsheet
(End-User View)
Reporting
(Executive View)
AI-powered
variables
Design
Model outputs
TEST
PROTO
-TYPE
15
Deliver
End-User EHR View
Real-Time ADE predictive Risk Integration
ACTIONABLE DATA INSIGHTS
Model screens every hospital admission patient ≥18
BUILD
16
Risk Focused Care
Model identifies patients at high-risk
for ADE at admission
Very high risk 6X more likely to be
consulted than low risk patients
MEASURE
3.2%
14.5%
17.7%
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
20%
Low High Very High
% Consulted
Top 10%*
(6,275)
*Real-time predictive score
AUC = 0.74
17
Deliver
MEASURE
2.7%
5.6%
5.9%
0.4%
8.9%
11.8%
0%
2%
4%
6%
8%
10%
12%
14%
% Consulted
Man vs. Machine
Man vs. Machine Man vs. Machine Man vs. Machine
LOW
(N=55,762)
HIGH
(N=3,086)
VERY HIGH
(N=3,189)
18
Man vs. Machine
MEASURE
Model Identified
Provider Initiated
GO-LIVE June 2017
Pre
Post
19
16%
12%
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
20%
Without Pharm Consult With Pharm Consult
30-Day Readmission Rate
Reduced Readmissions
P < 0.01
MEASURE
N= 4994*
N= 954*
*High/Very High Risk Patients
20
Model Impact
↑ Patient Safety
↓ 914 minor ADE
↓ 134 major ADE
↓ Readmissions by 25%
+
↓ ED visits
REAL-TIME
EHR-INTEGRATED
• 62,037 patients screened
• 6,275 Identified
• 2,775 Very High Risk
• 6X more likely ‘Very High Risk’
consults
• Saved 432 hours/FTE/year
• NNT=25
Resource
Utilization
Operational
Outcomes
Clinical
Outcomes
Cost
Savings
*AHRQ estimates based on cost of ADE prevented and ** HCUP_AHRQ estimates, through reduction in 30-day readmission rate
+
Relative reduction
SUSTAIN
$ 1.3 M savings* (ADE)
$ 38 K (savings in FTE hours)
$ 456 K savings ** (Readmissions)
Potential for $5.6 M per year
21
• End-user engagement
• Timely communication
• Proactive monitoring
• Seamless integration
• Scale and Sustain
Take Home Points
Teamwork makes Dreamwork!
22
Acknowledgements
Varun Sharma, PharmD
Lanora Gray, CPhT
Genae Sam, CPhT
Dr. Esmail Porsa, Chief Strategy Officer
Dr. Brett Moran, CMIO
Joe Longo, CIO
Julie Ly, PharmD (Epic Analyst)
Boryana Manz, PhD
Arun Nethi, Data Scientist
Lindsey Nace, Marketing and Communications
Philanthropic Innovation Grant Support
23
CONTACT INFORMATION
Manjula.Julka@pccinnovation.org Kristin.Alvarez@phhs.org
@manjula_julka
linkedin.com/in/manjula-julka-designdoc linkedin.com/in/kristin-snackey-alvarez-548a3718
Questions